def __init__(self, title=None, source_id=None, lang="en"):
super(File, self).__init__(title=title, source_id=source_id, lang=lang)
self.filename = get_name_from_url(source_id)
self.source_id = urljoin(FolkDCChef.BASE_URL, self.source_id)\
if source_id.startswith("/") else self.source_id
self.filepath = None
self.name = get_name_from_url_no_ext(self.filename)
def __init__(self, source_id, lang="en", name=None):
self.filename = get_name_from_url(source_id)
self.source_id = urljoin(
BASE_URL, source_id) if source_id.startswith("/") else source_id
self.filepath = None
self.lang = lang
self.name = "{}_{}".format(name, self.filename)
def __init__(self, source_id, lang="en", lincese="", name=None):
self.filename = get_name_from_url(source_id)
self.source_id = urljoin(
BASE_URL, source_id) if source_id.startswith("/") else source_id
self.filepath = None
self.lang = lang
self.name = "{}_{}".format(name, self.filename)
self.license = get_license(
licenses.CC_BY_NC_SA, copyright_holder=COPYRIGHT_HOLDER).as_dict()
def get_images(self, content):
for img in content.findAll("img"):
if img["src"].startswith("/"):
img_src = urljoin(BASE_URL, img["src"])
else:
img_src = img["src"]
filename = get_name_from_url(img_src)
if img_src not in self.images and img_src:
img["src"] = filename
self.images[img_src] = filename
def to_local_images(self, content):
images_urls = {}
for img in content.find_all("img"):
try:
img_src = img["src"]
except KeyError:
continue
else:
if img_src.startswith("/"):
img_src = urljoin(FolkDCChef.BASE_URL, img_src)
filename = get_name_from_url(img_src)
if img_src not in images_urls and img_src:
img["src"] = filename
images_urls[img_src] = filename
return images_urls
def __init__(self, title=None, source_id=None, lang="ar"):
super(File, self).__init__(title=title, source_id=source_id, lang=lang)
self.filename = get_name_from_url(source_id)
self.filepath = None
self.lang = lang